196
in four villages. Three of those admitted to hospital died about 31, 72, and 120 hours after ingesting amounts of seaweed described by family members as equivalent to half to one cereal bowl. Subsequent investigation identified ten other people (and one additional household) who had eaten seaweed from the same source. These individuals had only tasted the seaweed or had eaten only one or two spoonfuls. Six said they had not been affected and four reported symptoms such as loose stools or numbness of fingertips lasting a few hours. No patient reported reversal of hot-cold sensations. The seaweed eaten by all those affected was purchased from a single vendor at an open-air market that was open on Saturday and Sunday mornings only. Two vendors selling the seaweed Polycavernosa tsudai (formerly Gracilaria edulis) had collected their supply in the same area on the same day; between them they sold about 20 kg before remaining supplies were confiscated by health authorities. The affected seaweed was collected during high tide by loosening it with feet and toes and then allowing it to float to the surface. Unaffected seaweed was collected at low tide, which is the more usual practice of both vendors. Preliminary toxicological studies have identified a subtance with some characteristics of palytoxin in seaweed obtained from an affected household. Palytoxin is a virulent marine toxin reported in association with colonial anemones (Palythoa spp)/,2 fish,3and crabs.4 Additional studies are underway to further characterise the toxin, to delineate its distribution in the waters around Guam, and to find out how it came to contaminate normally edible seaweed. Because of the unusual nature of this outbreak we invite comments from anyone knowing of similar incidents. We thank Dr Robert Dickey, Dr Yoshitsugi Hokama, and Dr Takeshi Yasumoto, who have done preliminary studies of the toxin, and Susann Wilkins who identified the seaweed.
Department of Public Health and Social Services, PO Box 2816, Agana, Guam 96910 Guam Memorial
ROBERT L. HADDOCK
Hospital Authority,
OLIVIA L. T. CRUZ
Tamuning
1 Scheuer PJ The chemistry of toxins isolated from some marine organisms. Program Chem Org Nat Prod 1980; 22: 265-83. 2 Mosher HS, Fuhrman FA, Buckwald HD, Fisher HG. Tarichatoxin-tetrodotoxin a potent neurotoxin. Science 1964; 144: 1100-10 3. Noguchi T, Hwang D, Arakawa O, et al. Palytoxin as the causative agent in the parrotfish poisoning. In. Gopalaknshnakone P, Tan CK, ed. Progress in venom and toxin research: proceedings of the First Asia-Pacific Congress on Animal, Plant and Microbial Toxins (Singapore, June, 1987). Singapore National University of Singapore, 1987: 325-35. 4. Fukui M, Yasumura D, Murata M, et al. Occurrence of palytoxin in crabs and fish In. Gopalakrishnakone P, Tan CK, ed. Progress in venom and toxin research proceedings of the First Asia-Pacific Congress on Animal, Plant and Microbial Toxins (Singapore, June, 1987). Singapore. National University of Singapore, 1987: 477-82
Chorion villus
sampling
SiR,—The results of the Medical Research Council European Trial of chorion villus sampling (CVS) (June 22, p 1491) have been long awaited by those of us involved in prenatal diagnosis, unfortunately they are unlikely to provide the practising obstetrician with useful information that he or she can use in
counselling patients. The study began when CVSwas only just becoming established. Although there was an attempt to control for the experience of the operators, skill in this procedure depends not only on the number of cases but also with how often they are done. The attempts by the working party to stratify their data based on the early, mid, and late part of the study does not really address the issue of training and experience. Most operators accept that there is a fairly substantial learning curve, especially for the transcervical technique. Evaluation is made more difficult by a general drift from transcervical to transabdominal procedures-indeed, in recent years most have opted for the latter approach, which may carry less risk. It
comes as no surprise that CVS has a slightly greater risk of producing miscarriage than amniocentesis and most would counsel
along these lines. The numerical risk will depend on local skill, some units being better than others. One positive feature of the study is the reliability of CVS in terms of laboratory interpretation, which contrasts with much other published work. Department of Fetal Medicine, University of Birmingham, Birmingham Maternity Hospital, Birmingham B15 2TG, UK
M.
J. WHITTLE
SIR,-The Medical Research Council European trial and the Canadian study’ were large, prospective, and well randomised. Neither states when a clinician is proficient enough to do amniocentesis or CVS. In the MRC trial there was no change in performance with increasing experience. Although analyses suggested that differential effects were least amongst the final third of the sample recruited, this was not statistically significant. We have not allowed our patients to be exposed to the fetal mortality rate described since only experienced, trained obstetricians who have been through a planned programme do CVS or amniocentesis. CVS may not be for every consultant but should perhaps be the province of specialised centres. 17% of CVS procedures and 5% of amniocenteses were described in the European trial as "difficult". Such rates will not be the experience of expert departments. However, pregnancy outcomes were no worse for the difficult
procedures. We do not think that fetal losses should include all terminations because this will include terminations for social reasons. Furthermore, it is unfortunate that not all fetuses and babies that died had chromosome studies done. In the European trial there were fewer fetal deaths between 20 and 28 weeks and stillbirths before 32 weeks in the CVS group than in the amniocentesis group. There was also a better survival rate for babies born between 20 and 27 weeks in the CVS group (10/19) compared with the amniocentesis group (2/14). The higher neonatal death rate for the CVS group were due to more immature infants born before 32 weeks. This may merely reflect the neonatal services available. In centres where CVS was mainly by the transcervical route the rate difference in fetal loss was 3-9 and for centres preferring the transabdominal route it was 30. Centres unable to decide which procedure they were proficient in did both ("mixed centres") and had a much higher loss rate of 7’ 1. The MRC working-party suggests amniocentesis at 10-14 weeks as an attractive alternative to CV S for chromosomal and single gene defects, especially when rapid DNA amplification techniques are available. Unfortunately, the trial does not involve first-trimester amniocentesis and rapid DNA amplification is neither easy nor cheap. Until early amniocentesis has been evaluated in a randomised trial there are no grounds for suggesting this approach. Department of Obstetrics and Gynaecology, Faculty of Medicine, City Hospital, Nottingham NG5 1PB, UK
HENRY CHENG DAVID T. Y. LIU
Group. Multicentre randomised clinical trial of chonon villus sampling and amniocentesis. Lancet 1989, i 1-6.
1. Canadian Collaborative CVS-Amniocentesis Clinical Trial
Valproic acid and spina bifida SIR,-Your June 1 editorial states that the association between valproic acid (VA) and spina bifida (SB) is more common in some countries than in others. The power of detection of such association is related to the proportion of exposed people in the population: an association between defect and exposure can be better detected in populations with a high proportion of exposed people.The proportion of exposed people necessary to detect the association is even more important if the defect is rare, and this may be the situation for VA and SB. In the Rhone-Alps region of France,’ in Italy,2 and in Spain,3all populations where the prevalence of SB is less than 05 per 1000,’ the association was identified when the proportion of use of VA among epileptic mothers of malformed children was 20% or more. Thus, the lack of association in
197
presented with SB. We used the group of 119 malformed infants of epileptic mothers to see if the proportion of SB in this group correlates with the percentage of use of VA by their mothers over time. Fig 1 shows that the shape of the curve for the percentage of use of VA by this group of mothers is similar to that of the proportion of SB among the malformed infants of these mothers. We then plotted for each year the percentage of use of VA against percentage SB. In the years in which the use of VA by epileptic mothers who had a malformed child increased, the proportion of SB among the malformed infants of those mothers also increased (linear regression, p = 0-017). An earlier Lancet editorial asked for correlation of trends in SB
prevalence and sales of VA. This is not easy. The population of mothers potentially exposed to VA is not large, and the number of cases of SB that they could produce, even if at high risk, is not enough to produce a statistically significant increase in the prevalence of SB. In our data, the secular trend of the prevalence of SB does not differ over time, despite the fact that we have detected an association between use of VA and SB. 3.5 To study the correlation of use of VA and the production of SB it is necessary, as we have shown, to define the population that could be exposed and analyse data in that specific population. The correlation we found between VA use and proportion of SB among the malformed children of epileptic mothers, plus the fact Fig 1-Annual percentage of SB and of exposure to malformed children born to epileptic mothers.
VA among
populations with higher frequency of SB could be due to a very low proportion of people exposed to VA. You also stated that the evidence for the association could easily be influenced by the flow of negative case-reports, assuming that an infant prenatally exposed to an antiepileptic drug is more likely to come to medical attention than a non-exposed child. This is not so for the exposed and non-exposed children in the Spanish Collaborative Study of Congenital Malformations (ECEMC),5 a case-control study in which the collaborating physicians first identify the cases (children with major or minor congenital defects detected during the first three days of life) and select the controls, and only then interview the mothers. When they select cases and controls they do not know about prenatal exposure to anticonvulsant drugs, and our results on the association between VA and SB3sare not biased in the way your editorial suggests. From April, 1976, to September, 1990, we identified 119 malformed children whose mothers had epilepsy. 7 (5-9%) 25
7
that all the were
cases
of SB among this group of malformed children
prenatally exposed to VA, strongly supports a causal relation.
Supported in part by a grant from the Direccion General de Farmacia y Productos Sanitarios and by the Direccion General de Planificación Sanitaria, Ministerio de Sanidad y Consumo. I thank Dr Luis Prieto for help with the statistical analysis. ECEMC, Hospital Universitario San Carlos, Faculty of Medicine, Universidad Complutense, 28040, Madrid, Spain
MARÍA-LUISA MARTINEZ-FRIAS
1. Robert E, Guibaud P. Maternal valproic add and congenital neural tube defects. Lancet 1982; ii: 937. 2. Mastroiacovo P, Bertollini R, Morandini S, Segni G. Maternal epilepsy, valproate exposure, and birth defects. Lancet 1983; ii: 1499. 3. Martinez-Frías ML, Rodriguez Pinílla E, Salvador J. Valproate and spina bifida. Lancet 1989; i: 611-12. 4. International Clearinghouse for Birth Defects Monitoring Systems (ICBDMS). Annual report, 1987 5. Martínez-Frías ML. Clinical manifestation of prenatal exposure to valproic acid using case reports and epidemiologic information. Am J Med Genet 1991; 37: 277-82. 6. Editorial. Valproate and malformations. Lancet 1982; ii. 1314.
Near-fatal uterine perforation during transcervical endometrial resection SIR,-A 33-year-old woman being treated by transcervical resection of the endometrium (TCRE) for menorrhagia under general anaesthesia without relaxation suddenly and very forcefully adducted both her thighs. The procedure was stopped since it was thought to be complete. Uterine perforation was suspected when the patient complained of severe lower abdominal pain. Laparoscopy revealed a perforation in the area of the right uterine cornu, with bowel perforations and haemoperitoneum. At laparotomy multiple perforations of the terminal ileum and injury to the mesenteric and sacral vessels were found. The terminal ileum was resected, with an end-to-end anastomosis, and other perforations were repaired. During the operation the patient was critically ill with unrecordable blood pressure at times. Postoperatively she was in intensive care for 3 days. She had a second laparotomy for anastomotic stenosis after 14 days. She was discharged 26 days after her initial operation. Her menorrhagia has improved but she still requires dydrogesterone 10 mg three times a day. She has also had chronic pelvic pain and diarrhoea since the
operation.
Fig 2-Distribution of years by % of SB in malformed children of
epileptic mothers and % of exposure to VA
in their mothers.
TCRE has been suggested as the treatment of choice for menorrhagia for most patients, if surgery is required;1.2and complications are said to be infrequent.z3 Electrical stimulation of the obturator nerve during transurethral resection of bladder tumour is well known. It leads to violent adduction of the thighs and
in four villages. Three of those admitted to hospital died about 31, 72, and 120 hours after ingesting amounts of seaweed described by family members as equivalent to half to one cereal bowl. Subsequent investigation identified ten other people (and one additional household) who had eaten seaweed from the same source. These individuals had only tasted the seaweed or had eaten only one or two spoonfuls. Six said they had not been affected and four reported symptoms such as loose stools or numbness of fingertips lasting a few hours. No patient reported reversal of hot-cold sensations. The seaweed eaten by all those affected was purchased from a single vendor at an open-air market that was open on Saturday and Sunday mornings only. Two vendors selling the seaweed Polycavernosa tsudai (formerly Gracilaria edulis) had collected their supply in the same area on the same day; between them they sold about 20 kg before remaining supplies were confiscated by health authorities. The affected seaweed was collected during high tide by loosening it with feet and toes and then allowing it to float to the surface. Unaffected seaweed was collected at low tide, which is the more usual practice of both vendors. Preliminary toxicological studies have identified a subtance with some characteristics of palytoxin in seaweed obtained from an affected household. Palytoxin is a virulent marine toxin reported in association with colonial anemones (Palythoa spp)/,2 fish,3and crabs.4 Additional studies are underway to further characterise the toxin, to delineate its distribution in the waters around Guam, and to find out how it came to contaminate normally edible seaweed. Because of the unusual nature of this outbreak we invite comments from anyone knowing of similar incidents. We thank Dr Robert Dickey, Dr Yoshitsugi Hokama, and Dr Takeshi Yasumoto, who have done preliminary studies of the toxin, and Susann Wilkins who identified the seaweed.
Department of Public Health and Social Services, PO Box 2816, Agana, Guam 96910 Guam Memorial
ROBERT L. HADDOCK
Hospital Authority,
OLIVIA L. T. CRUZ
Tamuning
1 Scheuer PJ The chemistry of toxins isolated from some marine organisms. Program Chem Org Nat Prod 1980; 22: 265-83. 2 Mosher HS, Fuhrman FA, Buckwald HD, Fisher HG. Tarichatoxin-tetrodotoxin a potent neurotoxin. Science 1964; 144: 1100-10 3. Noguchi T, Hwang D, Arakawa O, et al. Palytoxin as the causative agent in the parrotfish poisoning. In. Gopalaknshnakone P, Tan CK, ed. Progress in venom and toxin research: proceedings of the First Asia-Pacific Congress on Animal, Plant and Microbial Toxins (Singapore, June, 1987). Singapore National University of Singapore, 1987: 325-35. 4. Fukui M, Yasumura D, Murata M, et al. Occurrence of palytoxin in crabs and fish In. Gopalakrishnakone P, Tan CK, ed. Progress in venom and toxin research proceedings of the First Asia-Pacific Congress on Animal, Plant and Microbial Toxins (Singapore, June, 1987). Singapore. National University of Singapore, 1987: 477-82
Chorion villus
sampling
SiR,—The results of the Medical Research Council European Trial of chorion villus sampling (CVS) (June 22, p 1491) have been long awaited by those of us involved in prenatal diagnosis, unfortunately they are unlikely to provide the practising obstetrician with useful information that he or she can use in
counselling patients. The study began when CVSwas only just becoming established. Although there was an attempt to control for the experience of the operators, skill in this procedure depends not only on the number of cases but also with how often they are done. The attempts by the working party to stratify their data based on the early, mid, and late part of the study does not really address the issue of training and experience. Most operators accept that there is a fairly substantial learning curve, especially for the transcervical technique. Evaluation is made more difficult by a general drift from transcervical to transabdominal procedures-indeed, in recent years most have opted for the latter approach, which may carry less risk. It
comes as no surprise that CVS has a slightly greater risk of producing miscarriage than amniocentesis and most would counsel
along these lines. The numerical risk will depend on local skill, some units being better than others. One positive feature of the study is the reliability of CVS in terms of laboratory interpretation, which contrasts with much other published work. Department of Fetal Medicine, University of Birmingham, Birmingham Maternity Hospital, Birmingham B15 2TG, UK
M.
J. WHITTLE
SIR,-The Medical Research Council European trial and the Canadian study’ were large, prospective, and well randomised. Neither states when a clinician is proficient enough to do amniocentesis or CVS. In the MRC trial there was no change in performance with increasing experience. Although analyses suggested that differential effects were least amongst the final third of the sample recruited, this was not statistically significant. We have not allowed our patients to be exposed to the fetal mortality rate described since only experienced, trained obstetricians who have been through a planned programme do CVS or amniocentesis. CVS may not be for every consultant but should perhaps be the province of specialised centres. 17% of CVS procedures and 5% of amniocenteses were described in the European trial as "difficult". Such rates will not be the experience of expert departments. However, pregnancy outcomes were no worse for the difficult
procedures. We do not think that fetal losses should include all terminations because this will include terminations for social reasons. Furthermore, it is unfortunate that not all fetuses and babies that died had chromosome studies done. In the European trial there were fewer fetal deaths between 20 and 28 weeks and stillbirths before 32 weeks in the CVS group than in the amniocentesis group. There was also a better survival rate for babies born between 20 and 27 weeks in the CVS group (10/19) compared with the amniocentesis group (2/14). The higher neonatal death rate for the CVS group were due to more immature infants born before 32 weeks. This may merely reflect the neonatal services available. In centres where CVS was mainly by the transcervical route the rate difference in fetal loss was 3-9 and for centres preferring the transabdominal route it was 30. Centres unable to decide which procedure they were proficient in did both ("mixed centres") and had a much higher loss rate of 7’ 1. The MRC working-party suggests amniocentesis at 10-14 weeks as an attractive alternative to CV S for chromosomal and single gene defects, especially when rapid DNA amplification techniques are available. Unfortunately, the trial does not involve first-trimester amniocentesis and rapid DNA amplification is neither easy nor cheap. Until early amniocentesis has been evaluated in a randomised trial there are no grounds for suggesting this approach. Department of Obstetrics and Gynaecology, Faculty of Medicine, City Hospital, Nottingham NG5 1PB, UK
HENRY CHENG DAVID T. Y. LIU
Group. Multicentre randomised clinical trial of chonon villus sampling and amniocentesis. Lancet 1989, i 1-6.
1. Canadian Collaborative CVS-Amniocentesis Clinical Trial
Valproic acid and spina bifida SIR,-Your June 1 editorial states that the association between valproic acid (VA) and spina bifida (SB) is more common in some countries than in others. The power of detection of such association is related to the proportion of exposed people in the population: an association between defect and exposure can be better detected in populations with a high proportion of exposed people.The proportion of exposed people necessary to detect the association is even more important if the defect is rare, and this may be the situation for VA and SB. In the Rhone-Alps region of France,’ in Italy,2 and in Spain,3all populations where the prevalence of SB is less than 05 per 1000,’ the association was identified when the proportion of use of VA among epileptic mothers of malformed children was 20% or more. Thus, the lack of association in
197
presented with SB. We used the group of 119 malformed infants of epileptic mothers to see if the proportion of SB in this group correlates with the percentage of use of VA by their mothers over time. Fig 1 shows that the shape of the curve for the percentage of use of VA by this group of mothers is similar to that of the proportion of SB among the malformed infants of these mothers. We then plotted for each year the percentage of use of VA against percentage SB. In the years in which the use of VA by epileptic mothers who had a malformed child increased, the proportion of SB among the malformed infants of those mothers also increased (linear regression, p = 0-017). An earlier Lancet editorial asked for correlation of trends in SB
prevalence and sales of VA. This is not easy. The population of mothers potentially exposed to VA is not large, and the number of cases of SB that they could produce, even if at high risk, is not enough to produce a statistically significant increase in the prevalence of SB. In our data, the secular trend of the prevalence of SB does not differ over time, despite the fact that we have detected an association between use of VA and SB. 3.5 To study the correlation of use of VA and the production of SB it is necessary, as we have shown, to define the population that could be exposed and analyse data in that specific population. The correlation we found between VA use and proportion of SB among the malformed children of epileptic mothers, plus the fact Fig 1-Annual percentage of SB and of exposure to malformed children born to epileptic mothers.
VA among
populations with higher frequency of SB could be due to a very low proportion of people exposed to VA. You also stated that the evidence for the association could easily be influenced by the flow of negative case-reports, assuming that an infant prenatally exposed to an antiepileptic drug is more likely to come to medical attention than a non-exposed child. This is not so for the exposed and non-exposed children in the Spanish Collaborative Study of Congenital Malformations (ECEMC),5 a case-control study in which the collaborating physicians first identify the cases (children with major or minor congenital defects detected during the first three days of life) and select the controls, and only then interview the mothers. When they select cases and controls they do not know about prenatal exposure to anticonvulsant drugs, and our results on the association between VA and SB3sare not biased in the way your editorial suggests. From April, 1976, to September, 1990, we identified 119 malformed children whose mothers had epilepsy. 7 (5-9%) 25
7
that all the were
cases
of SB among this group of malformed children
prenatally exposed to VA, strongly supports a causal relation.
Supported in part by a grant from the Direccion General de Farmacia y Productos Sanitarios and by the Direccion General de Planificación Sanitaria, Ministerio de Sanidad y Consumo. I thank Dr Luis Prieto for help with the statistical analysis. ECEMC, Hospital Universitario San Carlos, Faculty of Medicine, Universidad Complutense, 28040, Madrid, Spain
MARÍA-LUISA MARTINEZ-FRIAS
1. Robert E, Guibaud P. Maternal valproic add and congenital neural tube defects. Lancet 1982; ii: 937. 2. Mastroiacovo P, Bertollini R, Morandini S, Segni G. Maternal epilepsy, valproate exposure, and birth defects. Lancet 1983; ii: 1499. 3. Martinez-Frías ML, Rodriguez Pinílla E, Salvador J. Valproate and spina bifida. Lancet 1989; i: 611-12. 4. International Clearinghouse for Birth Defects Monitoring Systems (ICBDMS). Annual report, 1987 5. Martínez-Frías ML. Clinical manifestation of prenatal exposure to valproic acid using case reports and epidemiologic information. Am J Med Genet 1991; 37: 277-82. 6. Editorial. Valproate and malformations. Lancet 1982; ii. 1314.
Near-fatal uterine perforation during transcervical endometrial resection SIR,-A 33-year-old woman being treated by transcervical resection of the endometrium (TCRE) for menorrhagia under general anaesthesia without relaxation suddenly and very forcefully adducted both her thighs. The procedure was stopped since it was thought to be complete. Uterine perforation was suspected when the patient complained of severe lower abdominal pain. Laparoscopy revealed a perforation in the area of the right uterine cornu, with bowel perforations and haemoperitoneum. At laparotomy multiple perforations of the terminal ileum and injury to the mesenteric and sacral vessels were found. The terminal ileum was resected, with an end-to-end anastomosis, and other perforations were repaired. During the operation the patient was critically ill with unrecordable blood pressure at times. Postoperatively she was in intensive care for 3 days. She had a second laparotomy for anastomotic stenosis after 14 days. She was discharged 26 days after her initial operation. Her menorrhagia has improved but she still requires dydrogesterone 10 mg three times a day. She has also had chronic pelvic pain and diarrhoea since the
operation.
Fig 2-Distribution of years by % of SB in malformed children of
epileptic mothers and % of exposure to VA
in their mothers.
TCRE has been suggested as the treatment of choice for menorrhagia for most patients, if surgery is required;1.2and complications are said to be infrequent.z3 Electrical stimulation of the obturator nerve during transurethral resection of bladder tumour is well known. It leads to violent adduction of the thighs and
